Efficient storage of small random changes to data on disk

Abstract

The embodiments described herein provide a system and method for efficiently storing small, random modifications or changes to data on one or more storage devices, such as disks, of storage servers coupled to a host computer in a network environment. Illustratively, the data is stored in a region of a byte-addressable, persistent memory of the host computer and is replicated (i.e., copied) as changed data of the region on the disks at the granularity at which it was modified, e.g., at the byte-addressable granularity. To that end, each storage server employs a data structure (e.g., a Fibonacci array) that is configured to efficiently accumulate the small, random data changes into one or more large blocks of changed data for storage on the disks in a manner that realizes the streaming bandwidth of the disk.

Description

BACKGROUND[0001]1. Technical Field[0002]The present disclosure relates to storage of data and, more specifically, to efficient storage of small random changes to data on one or more disks coupled to a host computer in a network environment.[0003]2. Background Information[0004]Many modern computing algorithms are page-based and implemented in a kernel of an operating system executing on a host computer. Paging is a memory management function that facilitates storage and retrieval of data in blocks or “pages” to and from primary storage, such as disk. For example, assume that an application executing on the host computer utilizes a page-based algorithm to, e.g., insert a new node into a doubly-linked list. Execution of the algorithm may result in a first modified (“dirtied”) page, i.e., the page with a previous pointer, a second dirtied page, i.e., the page with a next pointer, and a third dirtied page containing the newly inserted node. Modification of the pages requires a number of ...

AI Technical Summary

Benefits of technology

The patent describes a system and method for efficiently storing small, random modifications or changes to data on one or more storage devices, such as disks, of storage servers coupled to a host computer in a network environment. The data is stored in a region of a byte-addressable, persistent memory of the host computer and is replicated as changed data on the disks at the granularity at which it was modified. The system employs a data structure, such as an on-disk array, that is organized as a series of levels and is parameterized by block size. The technical effect of this patent is to provide a more efficient way of storing small, random data changes on disks in a network environment, which can be vulnerable in the event of a disaster.

Problems solved by technology

Yet, even safe and consistent data stored in the persistent memory may be vulnerable in the event of a disaster because there is only a single copy of the data on the host computer.

However, disks generally provide good streaming bandwidth performance (e.g., reading and writing of a large number of sequential blocks or “track reads”) but do not perform well on small random accesses (i.e., reading and writing a single disk sector preceded by a disk seek).

In other words, disks operate most efficiently in sequential or streaming bandwidth mode, whereas small random accesses (such as the random seek operations described above) can substantially slow the performance of disks.

Images